Copeia 103, No. 3, 2015, 595–620 The Gill-Arch Musculature of Protanguilla, the Morphologically Most Primitive Eel (Teleostei: Anguilliformes), Compared with That of Other Putatively Primitive Extant Eels and Other Elopomorphs Victor G. Springer1 and G. David Johnson1 The gill-arch musculature and associated aspects of the skeleton of the anguilliforms Protanguilla, Conger, Anguilla, and the synaphobranchids Synaphobranchus and Simenchelys are described, illustrated, and compared. We identify nine anguilliform synapomorphies, seven myological and two osteological and all but two reported for the first time. We also describe one myological and one osteological synapomorphy of the Anguilliformes minus Protanguilla. Our study strongly corroborates the monophyly of the Anguilliformes, which has never been seriously challenged by morphological evidence, and is also supported by molecular analyses. Furthermore, it offers additional morphological support for placing Protanguilla as the sister group of all other eels. The condition of many gill-arch skeletal and muscular features are also treated in other elopomorphs and, occasionally, osteoglossomorphs and more primitive actinopterygians. For comparison, we also include the description of the gill-arch muscles of a specialized eel species of the family Serrivomeridae and discuss analyses of its closest family relationships as indicated by several molecular studies. and its anterior attachments, were removed. Techniques were There is no such thing as good or bad evidence; evidence is highly variable and often ad hoc. The muscles, many of which evidence. are quite delicate, are usually entangled with nerves, blood Donn Eric Rosen in an unrecorded conversation with VGS vessels, and sundry connective tissues that must be removed before study and illustration. In removing these extraneous OHNSON et al. (2012) described a new family, genus, tissues, occasional damage to some muscles occurred. On the and species of eel (Protanguillidae, Protanguilla, P. illustrations, damage was adjusted by using information from J palau), which they hypothesized is the sister group of the same muscle on the opposite side of the specimen, or in all other eels, both Recent and fossil. Their hypothesis was a few instances from other specimens that were undamaged based on Protanguilla’s possession of more plesiomorphic in the particular area (but were otherwise in poorer condition osteological characters than any other eel taxon, and on than the specimen being illustrated). Most of the muscle a cladistic analysis of the mitogenomic characters of damage involved the bilaterally paired ventral and dorsal Protanguilla and representatives of all other families of Recent retractors (VR, DR), which Nelson (1967a) reported are eels. As noted by Johnson et al., exclusive of Protanguilla, present in all eels. These retractors are anterior continuations ‘‘there is no morphology-based consensus of which Recent of the inner, longitudinal muscle layer of the sphincter oesophagi. Anteroventrally, the VR attaches to the autoge- eels are the most primitive,’’ but ‘‘based on gill-arch structure nous ventral tooth plate (TP5) and/or the associated fifth and other osteological features . [authors have suggested ceratobranchial (Cb5). Dorsally, each DR attaches to upper that] they would most probably be found among the families tooth plate 4 (UP4), and perhaps to pharyngobranchial 3. Synaphobranchidae, Congridae or Anguillidae . ’’ This Springer and Johnson (2004; henceforth, S&J) did not treat situation persists. In the present study, we examine a soft the dorsal esophageal retractors, and we do not comment on anatomical system, the gill-arch muscles, for evidence them herein. We devote little attention to the ventral bearing on Protanguilla’s anguilliform intra-relationships. retractors. Posterior to the attachments of the retractors, the Additionally, we comment on other characters and synapo- outer transverse or circular esophageal muscle layer sur- morphies that appear independently in other elopomorphs rounds the longitudinal layer. and less closely related teleostean taxa. For comparison, we Our descriptions and illustrations of anguilliform gill-arch muscles are mainly restricted to the same eel taxa treated by include the description of the gill-arch muscles of a special- S&J, who provided descriptions and illustrations of these ized eel species of the family Serrivomeridae, which three muscles for three of the four families variously purported to molecular studies have suggested is closely related to the represent the most plesiomorphic anguilliforms (Protanguil- Anguillidae: Johnson et al. (2012) and Tang and Fielitz lidae was unknown at that time). For those three anguilli- (2013), which placed Serrivomeridae and Anguillidae as sister forms, we reproduce S&J’s illustrations of the dorsal gill-arch groups, and Inoue et al. (2010), which placed Serrivomeridae muscles, with minor changes in labeling, but not the + Nemichthyidae as sister to the Anguillidae. descriptions, for which the reader is referred to S&J. We describe and illustrate the ventral gill-arch muscles for all these taxa, as well as both the dorsal and ventral gill-arch MATERIALS AND METHODS muscles of two additional genera, Protanguilla (Protanguilli- To study the gill-arch muscles, the entire gill-arch system and dae), described in 2012, and Simenchelys (Synaphobranchi- some associated bones and muscles, particularly the anterior dae), not treated by S&J. Additionally, we provide a complete portion of the sternohyoideus (not a ventral gill-arch muscle) description of the gill-arch musculature of a specialized eel, 1 Division of Fishes MRC 159, Department of Vertebrate Zoology, National Museum of Natural History, P.O. Box 37012, Washington, D.C. 20013-7012; E-mail: (VGS) [email protected]; and (GDJ) [email protected]. Send reprint requests to either VGS or GDJ. Submitted: 12 January 2015. Accepted: 1 March 2015. Associate Editor: G. Arratia. F 2015 by the American Society of Ichthyologists and Herpetologists DOI: 10.1643/CI-14-152 Published online: August 20, 2015 596 Copeia 103, No. 3, 2015 Fig. 1. Cladogram showing suprageneric taxa referred to in the present study extracted from the molecular study of Betancur et al. (2013), of which the lower part agrees with morphological evidence of Grande (2010) and the upper part with Arratia (1999, 2000a, 2000b, 2010, etc.). For clupeocephalan intrarelationships see Springer and Johnson (2004:fig. 3). Serrivomer beanii Gill and Ryder. Descriptions or indications indicated as E. saurus in S&J); Megalops cyprinoides (Brouss- of selected gill-arch muscles are also provided for various onet), USNM 350468 (erroneously indicated in S&J text as anguilliform and non-anguilliform elopomorph taxa we 350458). examined. The descriptions and illustrations of the dorsal gill-arch musculature of many of the non-anguilliform ANATOMICAL ABBREVIATIONS WITH DEFINITIONS elopomorph taxa are available in S&J. AND DISCUSSIONS We base our discussions on an abbreviated molecular phylogeny of actinopterygians as hypothesized by Betancur Abbreviations for names of muscles and skeletal elements et al. (2013), expanded to include more clades within its variously and occasionally extensively described and anno- proposed basal clades (Fig. 1). tated follow (parenthetical abbreviations are equivalents A summary of the relevant dorsal and ventral gill-arch used by Nelson, 1966a, 1967a). Among the annotations, we muscles of all the main anguilliform taxa we treat and an indicate the most strongly supported synapomorphies for example representing each of the main non-anguilliform dorsal gill-arch elements derived from optimizing characters elopomorph clades are provided in Table 1. on the cladogram in Figure 1. Many of these correspond to Institutional abbreviations follow Sabaj Pe´rez (2014). those hypothesized by S&J (2004:fig. 3), which were based Specimens of anguilliforms are listed at the beginning of on a different, exclusively morphology-based cladogram each species description. With a few noted exceptions, compiled from the literature (see S&J, 2004:16–18, for information on the gill-arch muscles of Moringua, Kau- a discussion of the various sources used to construct that pichthys, Uropterygius, and Gymnothorax are derived from cladogram). Anguilliform synapomorphies for some ventral Nelson (1967a). Some partial descriptions are also presented gill-arch muscles are also hypothesized based either on our for the gill-arch muscles and/or skeleton for which the gill- own observations or on published information. arch muscles are not illustrated. These include: Ophichthus AC4—accessory element (cartilage) at distal end of puncticeps (Kaup), Ophicthidae, USNM 431417, ca. 600 mm ceratobranchial 4. Traditionally, this cartilage has been TL; Muraenesox cinereus (Forsskal), Muraenesocidae, USNM treated as epibranchial 5 (e.g., S&J), but Carvalho et al. 431418, ca. 540 mm TL; and Hoplunnis tenuis Ginsburg, (2013) clearly demonstrated that in teleosts the cartilage Nettastomatidae, USNM 431416, ca. 530 mm TL. develops from the distal cartilaginous end of ceratobran- Specimens of non-anguilliform elopomorphs are as fol- chial 4 and has no association with the fifth gill arch. All lows (taxa described and illustrated in S&J are indicated in non-anguilliform Elopomorpha, except the albulid Albula, parentheses: Notacanthus chemnitzi Bloch, USNM 214342 examined by S&J, and the highly specialized notacanthid (S&J), USNM 214340, ca. 345 SL, USNM 214341,